Context Navigation

source: trunk/FACT++/src/rootifysql.cc@ 19411

Visit:

Last change on this file since 19411 was 19403, checked in by tbretz, 7 years ago
include 'random' was missing for Peppermint 9 (gcc 7)
File size: 38.4 KB

Line
1	#include "Database.h"
2
3	#include <random>
4
5	#include <boost/regex.hpp>
6	#include <boost/tokenizer.hpp>
7	#include <boost/algorithm/string.hpp>
8	#include <boost/algorithm/string/join.hpp>
9
10	#include "tools.h"
11	#include "Time.h"
12	#include "Configuration.h"
13
14	#include <TROOT.h>
15	#include <TSystem.h>
16	#include <TFile.h>
17	#include <TTree.h>
18
19	using namespace std;
20
21	// ------------------------------------------------------------------------
22
23	void SetupConfiguration(Configuration &conf)
24	{
25	po::options_description control("Database options");
26	control.add_options()
27	("uri,u", var<string>()->required(), "Database link as in\n\tuser:password@server[:port]/database[?compress=0\|1].")
28	("query,q", var<string>(""), "MySQL query (overwrites --file)")
29	("file", var<string>("rootify.sql"), "An ASCII file with the MySQL query (overwrites --query)")
30	("ignore-null,i", po_switch(), "Do not skip rows containing any NULL field")
31	("display,d", po_switch(), "Displays contents on the screen (most usefull in combination with mysql statements as SHOW or EXPLAIN)")
32	("explain", po_switch(), "Requests an EXPLAIN from the server (shows the server optimized query)\nsee also https://dev.mysql.com/doc/refman/explain-output.html")
33	("profiling", po_switch(), "Turn on profiling and print profile")
34	("var.*", var<string>(), "Predefined SQL user variables (@VAR)")
35	("env.*", vars<string>(), "Predefined environment for substitutions in the query ($ENV)")
36	("list.*", var<string>(), "Predefined environment for substitutions in the query ($ENV). The list is read from the given file (one list entry per line)")
37	("print-connection", po_switch(), "Print database connection information")
38	("verbose,v", var<uint16_t>(1), "Verbosity (0: quiet, 1: default, 2: more, 3, ...)")
39	;
40
41	po::options_description ascii("ASCII output");
42	ascii.add_options()
43	("write,w", var<string>(""), "Write output to an ascii file")
44	("delimiter", var<string>(), "The delimiter used if contents are displayed with --display (default=\\t)")
45	("copy-shabang", po_switch(), "Copy the sha-bang line if exists to the output file")
46	("copy-header", po_switch(), "Copy the header (all line starting with '#' up to the first non-comment line to the output file")
47	("copy-query", po_switch(), "Copy the query to the ascii output file")
48	("copy-comments", po_switch(), "Copy all lines starting with '#' to the output file which are not part of header")
49	("copy-all", po_switch(), "An alias for --copy-header --copy-query --copy-comments")
50	;
51
52	po::options_description root("Root file options");
53	root.add_options()
54	("out,o", var<string>("rootify.root"), "Output root file name")
55	("force,f", po_switch(), "Force overwriting an existing root file ('RECREATE')")
56	("update", po_switch(), "Update an existing root file with the new tree ('UPDATE')")
57	("compression,c", var<uint16_t>(1), "zlib compression level for the root file")
58	("tree,t", var<string>("Result"), "Name of the root tree")
59	("ignore", vars<string>(), "Ignore the given columns")
60	("null,n", po_switch(), "Redirect the output file to /dev/null (mainly for debugging purposes, e.g. performance studies)")
61	("no-fill", po_switch(), "Do not fill events into the root file (mainly for debugging purposes, e.g. performance studies)")
62	;
63
64	po::options_description split("Splitting options");
65	split.add_options()
66	("split-sequence,S", vars<uint16_t>(), "Split data sequentially into several trees/files (e.g. 1, 1, 2)")
67	("split-quantile,Q", vars<double>(), "Split data randomly into several trees/files (e.g. 0.5, 1)")
68	("seed", var<uint64_t>(mt19937_64::default_seed), "Seed value in case of random split")
69	;
70
71	po::positional_options_description p;
72	p.add("file", 1); // The 1st positional options (n=1)
73	p.add("out", 1); // The 2nd positional options (n=1)
74
75	conf.AddOptions(control);
76	conf.AddOptions(ascii);
77	conf.AddOptions(root);
78	conf.AddOptions(split);
79	conf.SetArgumentPositions(p);
80	}
81
82	void PrintUsage()
83	{
84	cout <<
85	"rootifysql - Converts the result of a mysql query into a root file\n"
86	"\n"
87	"For convenience, this documentation uses the extended version of the options, "
88	"refer to the output below to get the abbreviations.\n"
89	"\n"
90	"Writes the result of a mysql query into a root file. For each column, a branch is "
91	"created of type double with the field name as name. This is usually the column name "
92	"if not specified otherwise by the AS mysql directive.\n"
93	"\n"
94	"Columns with CHAR or VARCHAR as field type are ignored. DATETIME, DATE and TIME "
95	"columns are converted to unix time (time_t). Rows containing any file which is "
96	"NULL are skipped if not suppressed by the --ignore-null option. Ideally, the query "
97	"is compiled in a way that no NULL field is returned. With the --display option the "
98	"result of the request is printed on the screen (NULL skipping still in action). "
99	"This can be useful to create an ascii file or to show results as 'SHOW DATABASES' "
100	"or 'EXPLAIN table'. To redirect the contents into an ascii file, the option -v0 "
101	"is useful. To suppress writing to an output file --null can be used.\n"
102	"\n"
103	"The default is to read the query from a file called rootify.sql. Except if a different "
104	"filename is specified by the --file option or a query is given with --query.\n"
105	"\n"
106	"As a trick, the rootify.sql file can be made excutable (chmod u+x rootify.sql). "
107	"If the first line contains '#!rootifysql', the script can be executed directly.\n"
108	"\n"
109	"Columns whose name start with @ are skipped. If you want them in your output file "
110	"give them a name using AS, e.g. 'SELECT @A:=5 AS A'.\n"
111	"\n"
112	"You can use variables in your sql query like @MyVar and define them on the "
113	"command line. In this example with --var.MyVar=5\n"
114	"\n"
115	"You can use environment definitions for substitutions in your SQL query. "
116	"For example --env.TEST=5 would replace $TEST or ${TEST} in your query by 5."
117	"If you specify one environment variable more than once, a list is created. "
118	"For example --env.TEST=1 --env.TEST=2 --env.TEST=3 would substitute "
119	"$TEST or ${TEST} by '1, 2, 3'. This is useful for the SQL `IN` keyword. "
120	"You can also read the values for an enviroment substitution from a file "
121	"(one element per line), e.g. --env.TEST=file.txt. Empty lines and lines "
122	"starting with a # are skipped.\n"
123	"\n"
124	"Comments in the query-file can be placed according to the SQL standard inline "
125	"/comment/ or introduced with # (shell script style) or -- (SQL style).\n"
126	"\n"
127	"For several purposes, it might be convenient to split the output to several "
128	"different root-trees or ascii files. This can be done using the --split-sequence (-S) "
129	"and the --split-quantile (-Q) options. If a split sequence is defined as "
130	"-S 1 -S 2 -S 1 the events are split by 1:2:1 in this sequence order. If "
131	"quantiled are given as -Q 0.5 -Q 0.6, the first tree will contain 50% of "
132	"the second one 10% and the third one 40%. The corresponding seed value can "
133	"be set with --seed.\n"
134	"\n"
135	"In case of success, 0 is returned, a value>0 otherwise.\n"
136	"\n"
137	"Usage: rootifysql [rootify.sql [rootify.root]] [-u URI] [-q query\|-f file] [-i] [-o out] [-f] [-cN] [-t tree] [-vN]\n"
138	"\n"
139	;
140	cout << endl;
141	}
142
143	struct ExplainParser
144	{
145	string sql;
146
147	vector<string> vec;
148
149	string substitute(string _str, const boost::regex &expr)
150	{
151	boost::smatch match;
152	while (boost::regex_search(_str, match, expr, boost::regex_constants::format_first_only))
153	{
154	const auto &len = match.length();
155	const auto &pos = match.position();
156	const auto &str = match.str();
157
158	const auto it = find(vec.cbegin(), vec.cend(), str);
159	const size_t id = it==vec.cend() ? vec.size() : it-vec.cbegin();
160
161	_str.replace(pos, len, "{"+to_string(id)+"}");
162
163	if (it==vec.cend())
164	vec.push_back(str);//.substr(1, str.size()-2));
165	}
166
167	return _str;
168	}
169
170	string substitute(const string &str, const string &expr)
171	{
172	return substitute(str, boost::regex(expr));
173	}
174
175	vector<string> queries;
176
177	string resub(string str)
178	{
179	// search for "KEYWORD expression"
180	boost::regex reg("\\{[0-9]+\\}");
181
182	boost::smatch match;
183	while (boost::regex_search(str, match, reg, boost::regex_constants::format_first_only))
184	{
185	const auto &len = match.length();
186	const auto &pos = match.position();
187	const auto &arg = match.str(); // Argument
188
189	const auto idx = atoi(arg.c_str()+1);
190
191	str.replace(pos, len, resub(vec[idx]));
192	}
193
194	return str;
195	}
196
197	void expression(string expr, size_t indent=0)
198	{
199	if (expr[0]=='{')
200	{
201	const auto idx = atoi(expr.c_str()+1);
202
203	// This is a subquery
204	if (vec[idx].substr(0,3)=="(/*")
205	{
206	cout << setw(indent) << ' ' << "(\n";
207	find_tokens(vec[idx], indent+4);
208	cout << setw(indent) << ' ' << ") ";
209	}
210	else
211	// This is just something to substitute back
212	if (vec[idx].substr(0,2)=="({")
213	{
214	cout << setw(indent) << ' ' << "(" << resub(vec[idx]) << ") ";
215	}
216	else
217	{
218	if (indent>0)
219	cout << setw(indent) << ' ';
220	cout << resub(vec[idx]);
221	}
222	}
223	else
224	{
225	if (indent>0)
226	cout << setw(indent) << ' ';
227	cout << resub(expr);
228	}
229	}
230
231	void find_tokens(string str, size_t indent=0)
232	{
233	// ( COMMENT )?( TOKEN )?(( {NNN} \| NNN )( AS\|ON ( {NNN}) ))?(,)?)
234	//regex reg("(\\/\\\\ select\\#[0-9]+\\ \\\\/\\ *)?([a-zA-Z ]+)?((\\{[0-9]+\\}\|[0-9]+)(\\ ?([Aa][Ss]\|[Oo][Nn])\\ ?(\\{[0-9]+\\}))?(,)?)");
235
236	const string _com = "\\/\\\\ select\\#[0-9]+\\ \\\\/\\ *";
237
238	const string _tok = "[a-zA-Z_ ]+";
239
240	const string _nnn = "\\{[0-9]+\\}\|[0-9]+";
241
242	const string _as = "\\ ?([Aa][Ss])\\ ?";
243
244	// ( _nnn ) ( _as ( _nnn ))?(,)? // can also match noting in between two {NNN}
245	const string _exp = "("+_nnn+")" + "("+_as+"("+_nnn+"))?(,)?";
246
247	// Matche: ( _com )? ( ( _tok )? ( _exp ) \| ( _tok ) )
248	boost::regex reg("("+_com+")?" + "(" + "("+_tok+")?"+"("+_exp+")" + "\|" + "("+_tok+")" + ")");
249
250	boost::smatch match;
251	while (boost::regex_search(str, match, reg, boost::regex_constants::format_first_only))
252	{
253
254	const auto &com = match.str(1); // comment
255	const auto &tok1 = Tools::Trim(match.str(3)); // token with expression
256	const auto &arg1 = match.str(5); // argument 1
257	const auto &as = match.str(7); // as
258	const auto &arg2 = match.str(8); // argument 2
259	const auto &comma = match.str(9); // comma
260	const auto &tok2 = Tools::Trim(match.str(10)); // token without expression
261
262	if (!com.empty())
263	cout << setw(indent) << ' ' << "\033[34m" << com << "\033[0m" << '\n';
264
265	if (!tok1.empty())
266	cout << setw(indent) << ' ' << "\033[32m" << tok1 << "\033[0m" << '\n';
267	if (!tok2.empty())
268	cout << setw(indent) << ' ' << "\033[32m" << tok2 << "\033[0m" << '\n';
269
270	if (!arg1.empty())
271	{
272	expression(arg1, indent+4);
273
274	if (!as.empty())
275	cout << " \033[33m" << as << "\033[0m ";
276
277	if (!arg2.empty())
278	expression(arg2);
279
280	if (!comma.empty())
281	cout << ',';
282
283	cout << '\n';
284	}
285
286	str = str.substr(match.position()+match.length());
287	}
288	}
289
290
291	ExplainParser(const string &_sql) : sql(_sql)
292	{
293	// substitute all strings
294	sql = substitute(sql, "'[^']*'");
295
296	// substitute all escaped sequences (`something`.`something-else`)
297	sql = substitute(sql, "`[^`]`(\\.`[^`]`)*");
298
299	// substitute all paranthesis
300	sql = substitute(sql, "[a-zA-Z0-9_]\\([^\\(\\)]\\)");
301
302	//cout << sql << "\n\n";
303	find_tokens(sql);
304	cout << endl;
305	}
306	};
307
308	// Remove queries...
309	void format(string sql)
310	{
311	ExplainParser p(sql);
312
313	/*
314
315	SELECT
316	[ALL \| DISTINCT \| DISTINCTROW ]
317	[HIGH_PRIORITY]
318	[STRAIGHT_JOIN]
319	[SQL_SMALL_RESULT] [SQL_BIG_RESULT] [SQL_BUFFER_RESULT]
320	[SQL_CACHE \| SQL_NO_CACHE] [SQL_CALC_FOUND_ROWS]
321	select_expr [, select_expr ...]
322	[FROM table_references
323	[PARTITION partition_list]
324	[WHERE where_condition]
325	[GROUP BY {col_name \| expr \| position}, ... [WITH ROLLUP]]
326	[HAVING where_condition]
327	[WINDOW window_name AS (window_spec)
328	[, window_name AS (window_spec)] ...]
329	[ORDER BY {col_name \| expr \| position}
330	[ASC \| DESC], ... [WITH ROLLUP]]
331	[LIMIT {[offset,] row_count \| row_count OFFSET offset}]
332	[INTO OUTFILE 'file_name'
333	[CHARACTER SET charset_name]
334	export_options
335	\| INTO DUMPFILE 'file_name'
336	\| INTO var_name [, var_name]]
337	[FOR {UPDATE \| SHARE} [OF tbl_name [, tbl_name] ...] [NOWAIT \| SKIP LOCKED]
338	\| LOCK IN SHARE MODE]]
339	*/
340
341	/*
342	table_references:
343	escaped_table_reference [, escaped_table_reference] ...
344
345	escaped_table_reference:
346	table_reference
347	\| { OJ table_reference }
348
349	table_reference:
350	table_factor
351	\| join_table
352
353	table_factor:
354	tbl_name [PARTITION (partition_names)]
355	[[AS] alias] [index_hint_list]
356	\| table_subquery [AS] alias [(col_list)]
357	\| ( table_references )
358
359	join_table:
360	table_reference [INNER \| CROSS] JOIN table_factor [join_condition]
361	\| table_reference STRAIGHT_JOIN table_factor
362	\| table_reference STRAIGHT_JOIN table_factor ON conditional_expr
363	\| table_reference {LEFT\|RIGHT} [OUTER] JOIN table_reference join_condition
364	\| table_reference NATURAL [INNER \| {LEFT\|RIGHT} [OUTER]] JOIN table_factor
365
366	join_condition:
367	ON conditional_expr
368	\| USING (column_list)
369
370	index_hint_list:
371	index_hint [, index_hint] ...
372
373	index_hint:
374	USE {INDEX\|KEY}
375	[FOR {JOIN\|ORDER BY\|GROUP BY}] ([index_list])
376	\| IGNORE {INDEX\|KEY}
377	[FOR {JOIN\|ORDER BY\|GROUP BY}] (index_list)
378	\| FORCE {INDEX\|KEY}
379	[FOR {JOIN\|ORDER BY\|GROUP BY}] (index_list)
380
381	index_list:
382	index_name [, index_name] ...
383	*/
384
385	}
386
387	int finish(Database &connection, const uint16_t &verbose, const bool &profiling, const bool &print_connection)
388	{
389	if (verbose>0)
390	{
391	try
392	{
393	const auto resw =
394	connection.query("SHOW WARNINGS").store();
395
396	if (resw.num_rows()>0)
397	cout << "\n" << resw.num_rows() << " Warning(s) issued:\n\n";
398
399	for (size_t i=0; i<resw.num_rows(); i++)
400	{
401	const mysqlpp::Row &roww = resw[i];
402
403	cout << roww["Level"] << '[' << roww["Code"] << "]: ";
404	cout << roww["Message"] << '\n';
405	}
406	cout << endl;
407
408	}
409	catch (const exception &e)
410	{
411	cerr << "\nSHOW WARNINGS\n\n";
412	cerr << "SQL query failed:\n" << e.what() << endl;
413	return 1;
414	}
415	}
416
417	if (profiling)
418	{
419	try
420	{
421	const auto N =
422	connection.query("SHOW PROFILES").store().num_rows();
423
424	const auto resp =
425	connection.query("SHOW PROFILE ALL FOR QUERY "+to_string(verbose?N-1:N)).store();
426
427	cout << '\n';
428	cout << left;
429	cout << setw(26) << "Status" << ' ';
430	cout << right;
431	cout << setw(11) << "Duration" << ' ';
432	cout << setw(11) << "CPU User" << ' ';
433	cout << setw(11) << "CPU System" << '\n';
434	cout << "--------------------------------------------------------------\n";
435	for (size_t i=0; i<resp.num_rows(); i++)
436	{
437	const mysqlpp::Row &rowp = resp[i];
438
439	cout << left;
440	cout << setw(26) << rowp["Status"] << ' ';
441	cout << right;
442	cout << setw(11) << rowp["Duration"] << ' ';
443	cout << setw(11) << rowp["CPU_user"] << ' ';
444	cout << setw(11) << rowp["CPU_system"] << '\n';
445	}
446	cout << "--------------------------------------------------------------\n";
447	cout << endl;
448	}
449	catch (const exception &e)
450	{
451	cerr << "\nSHOW PROFILE ALL\n\n";
452	cerr << "SQL query failed:\n" << e.what() << '\n' <<endl;
453	return 2;
454	}
455	}
456
457	if (print_connection)
458	{
459	try
460	{
461	// Exchange _send and _received as it is the view of the server
462	const auto &res1 = connection.query("SHOW STATUS LIKE 'Bytes_%'").store();
463	cout << left << setw(16) << res1[1]["Variable_name"] << ' ' << Tools::Scientific(res1[0]["Value"]) << endl;
464	cout << left << setw(16) << res1[0]["Variable_name"] << ' ' << Tools::Scientific(res1[1]["Value"]) << endl;
465	cout << endl;
466	}
467	catch (const exception &e)
468	{
469	cerr << "\nSHOW STATUS LIKE 'Bytes_%'\n\n";
470	cerr << "SQL query failed:\n" << e.what() << endl;
471	return 3;
472	}
473	}
474
475	if (verbose>0)
476	cout << "Success!\n" << endl;
477	return 0;
478
479	}
480
481	int main(int argc, const char* argv[])
482	{
483	Time start;
484
485	gROOT->SetBatch();
486
487	Configuration conf(argv[0]);
488	conf.SetPrintUsage(PrintUsage);
489	SetupConfiguration(conf);
490
491	if (!conf.DoParse(argc, argv))
492	return 127;
493
494	// ----------------------------- Evaluate options --------------------------
495	const string uri = conf.Get<string>("uri");
496	const string out = conf.Get<string>("out");
497	const string file = conf.Get<string>("file");
498	const string tree = conf.Get<string>("tree");
499	const bool force = conf.Get<bool>("force");
500	const bool ignorenull = conf.Get<bool>("ignore-null");
501	const bool update = conf.Get<bool>("update");
502	const bool display = conf.Get<bool>("display");
503	const string write = conf.Get<string>("write");
504	const bool noout = conf.Get<bool>("null");
505	const bool nofill = conf.Get<bool>("no-fill");
506	const bool explain = conf.Get<bool>("explain");
507	const bool profiling = conf.Get<bool>("profiling");
508	const uint16_t verbose = conf.Get<uint16_t>("verbose");
509	const uint16_t compression = conf.Get<uint16_t>("compression");
510	const string delimiter = conf.Has("delimiter") ? conf.Get<string>("delimiter") : "\t";
511
512	const bool copy_all = conf.Get<bool>("copy-all");
513	const bool copy_shabang = conf.Get<bool>("copy-shabang");
514	const bool copy_header = copy_all \|\| conf.Get<bool>("copy-header");
515	const bool copy_query = copy_all \|\| conf.Get<bool>("copy-query");
516	const bool copy_comments = copy_all \|\| conf.Get<bool>("copy-comments");
517
518	const vector<string> _ignore = conf.Vec<string>("ignore");
519	const bool print_connection = conf.Get<bool>("print-connection");
520	//const vector<Map> mymap = conf.Vec<Map>("map");
521
522	// ----------------------- Setup splitting ---------------------------------
523
524	vector<uint16_t> split_seq = conf.Vec<uint16_t>("split-sequence");
525	vector<double> split_quant = conf.Vec<double>("split-quantile");
526
527	if (!split_seq.empty() && !split_quant.empty())
528	throw runtime_error("Only splitting by --split-sequence or --split-quantile is allowed.");
529
530	const size_t num_split = ::max(split_seq.size(), split_quant.size()+1);
531
532	map<size_t, size_t> split_lut;
533	for (size_t i=0; i<split_seq.size(); i++)
534	{
535	const size_t sz = split_lut.size();
536	for (size_t j=0; j<split_seq[i]; j++)
537	split_lut.emplace(j+sz, i);
538	}
539
540	for (size_t i=0; i<split_quant.size(); i++)
541	if (split_quant[i]<0 \|\| split_quant[i]>=1)
542	throw runtime_error("Splitting quantiles must be in the range [0;1)");
543
544	for (size_t i=1; i<split_quant.size(); i++)
545	{
546	if (split_quant[i]<=split_quant[i-1])
547	throw runtime_error("Splitting quantiles must be in increasing order.");
548	}
549
550	// -------------------------------------------------------------------------
551
552	const auto vars = conf.GetWildcardOptions("var.*");
553
554	vector<string> variables;
555	for (auto var=vars.cbegin(); var!=vars.cend(); var++)
556	variables.emplace_back('@'+var->substr(4)+":="+Tools::Trim(conf.Get<string>(*var)));
557
558	// -------------------------------------------------------------------------
559
560	if (verbose>0)
561	{
562	cout << "\n------------------------ Rootify SQL -------------------------" << endl;
563	cout << "Start Time: " << Time::sql << Time(Time::local) << endl;
564	}
565
566	string query = conf.Get<string>("query");
567	if (query.empty())
568	{
569	if (verbose>0)
570	cout << "Reading query from file '" << file << "'." << endl;
571
572	ifstream fin(file);
573	if (!fin)
574	{
575	cerr << "Could not open query in '" << file << "': " << strerror(errno) << endl;
576	return 4;
577	}
578
579	getline(fin, query, (char)fin.eof());
580	}
581
582	if (query.empty())
583	{
584	cerr << "No query specified." << endl;
585	return 5;
586	}
587
588	// -------------------------------------------------------------------------
589
590	map<string, vector<string>> envs;
591
592	const auto &envs1 = conf.GetWildcardOptions("env.*");
593	for (auto env=envs1.cbegin(); env!=envs1.cend(); env++)
594	envs[env->substr(4)] = conf.Vec<string>(*env);
595
596	const auto &envs2 = conf.GetWildcardOptions("list.*");
597	for (auto env=envs2.cbegin(); env!=envs2.cend(); env++)
598	{
599	const string fname = conf.Get<string>(*env);
600	const string &ident = env->substr(5);
601
602	ifstream fin(fname);
603	if (!fin)
604	{
605	cerr << "Could not open environment in '" << fname << "' for ${" << ident << "}: " << strerror(errno) << endl;
606	return 6;
607	}
608	for (string line; getline(fin, line); )
609	{
610	const auto &l = Tools::Trim(line);
611	if (!l.empty() && l[0]!='#')
612	envs[ident].push_back(line);
613	}
614
615	if (verbose>0)
616	cout << "Found " << envs[ident].size() << " list element(s) for ${" << ident << "}" << endl;
617	}
618
619	for (auto env=envs.cbegin(); env!=envs.cend(); env++)
620	{
621	boost::regex rexpr("\\$(\\{"+env->first+"\\}\|"+env->first+"\\b)");
622	query = boost::regex_replace(query, rexpr, boost::join(env->second, ", "));
623	}
624
625	// -------------------------- Check for file permssion ---------------------
626	// Strictly speaking, checking for write permission and existance is not necessary,
627	// but it is convenient that the user does not find out that it failed after
628	// waiting long for the query result
629	//
630
631	// I am using root here instead of boost to be
632	// consistent with the access pattern by TFile
633	TString path(noout?"/dev/null":out.c_str());
634	gSystem->ExpandPathName(path);
635
636	if (!noout)
637	{
638	FileStat_t stat;
639	const Int_t exist = !gSystem->GetPathInfo(path, stat);
640	const Bool_t _write = !gSystem->AccessPathName(path, kWritePermission) && R_ISREG(stat.fMode);
641
642	if ((update && !exist) \|\| (update && exist && !_write) \|\| (force && exist && !_write))
643	{
644	cerr << "File '" << path << "' is not writable." << endl;
645	return 7;
646	}
647
648	if (!update && !force && exist)
649	{
650	cerr << "File '" << path << "' already exists." << endl;
651	return 8;
652	}
653	}
654
655	Time start2;
656
657	// --------------------------- Connect to database -------------------------------------------------
658
659	if (*query.rbegin()!='\n')
660	query += '\n';
661
662	if (verbose>0)
663	{
664	cout << "Connecting to database...\n";
665	cout << "Client Version: " << mysqlpp::Connection().client_version() << endl;
666	}
667
668	Database connection(uri); // Keep alive while fetching rows
669
670	if (verbose>0)
671	cout << "Server Version: " << connection.server_version() << endl;
672
673	if (print_connection)
674	{
675	try
676	{
677	const auto &res1 = connection.query("SHOW STATUS LIKE 'Compression'").store();
678	cout << "Compression of database connection is " << string(res1[0][1]) << endl;
679
680	const auto &res2 = connection.query("SHOW STATUS LIKE 'Ssl_cipher'").store();
681	cout << "Connection to databases is " << (string(res2[0][1]).empty()?"UNENCRYPTED":"ENCRYPTED ("+string(res2[0][1])+")") << endl;
682	}
683	catch (const exception &e)
684	{
685	cerr << "\nSHOW STATUS LIKE 'Compression'\n\n";
686	cerr << "SQL query failed:\n" << e.what() << endl;
687	return 9;
688	}
689	}
690
691	try
692	{
693	if (profiling)
694	connection.query("SET PROFILING=1").execute();
695	}
696	catch (const exception &e)
697	{
698	cerr << "\nSET profiling=1\n\n";
699	cerr << "SQL query failed:\n" << e.what() << endl;
700	return 10;
701	}
702
703	// -------------------------- Set user defined variables -------------------
704	if (variables.size()>0)
705	{
706	if (verbose>0)
707	cout << "Setting user defined variables..." << endl;
708
709	const string varset =
710	"SET\n "+boost::algorithm::join(variables, ",\n ");
711
712	try
713	{
714	connection.query(varset).execute();
715	}
716	catch (const exception &e)
717	{
718	cerr << '\n' << varset << "\n\n";
719	cerr << "SQL query failed:\n" << e.what() << endl;
720	return 11;
721	}
722
723	if (verbose>2)
724	cout << '\n' << varset << '\n' << endl;
725	}
726
727	// ------------------------- Explain query if requested --------------------
728
729	if (explain)
730	{
731	try
732	{
733	const auto res0 =
734	connection.query("EXPLAIN FORMAT=JSON "+query).store();
735
736	cout << res0[0][0] << endl;
737	cout << endl;
738
739	const mysqlpp::StoreQueryResult res1 =
740	connection.query("EXPLAIN "+query).store();
741
742	for (size_t i=0; i<res1.num_rows(); i++)
743	{
744	const mysqlpp::Row &row = res1[i];
745
746	cout << "\nid : " << row["id"];
747	cout << "\nselect type : " << row["select_type"];
748
749	if (!row["table"].is_null())
750	cout << "\ntable : " << row["table"];
751
752	if (!row["partitions"].is_null())
753	cout << "\npartitions : " << row["partitions"];
754
755	if (!row["key"].is_null())
756	cout << "\nselected key : " << row["key"] << " [len=" << row["key_len"] << "] out of (" << row["possible_keys"] << ")";
757
758	if (!row["type"].is_null())
759	cout << "\njoin type : " << row["type"];
760
761	//if (!row["possible_keys"].is_null())
762	// cout << "\npossible_keys: " << row["possible_keys"];
763
764	//if (!row["key_len"].is_null())
765	// cout << "\nkey_len : " << row["key_len"];
766
767	if (!row["ref"].is_null())
768	cout << "\nref : (" << row["ref"] << ") compared to the index";
769
770	if (!row["rows"].is_null())
771	cout << "\nrows : " << row["rows"];
772
773	if (!row["filtered"].is_null())
774	cout << "\nfiltered : " << row["filtered"];
775
776	if (!row["extra"].is_null())
777	cout << "\nExtra : " << row["extra"];
778
779	cout << endl;
780	}
781
782	cout << endl;
783
784	const mysqlpp::StoreQueryResult res2 =
785	connection.query("SHOW WARNINGS").store();
786
787	for (size_t i=0; i<res2.num_rows(); i++)
788	{
789	const mysqlpp::Row &row = res2[i];
790
791	// 1003 //
792	cout << row["Level"] << '[' << row["Code"] << "]:\n";
793	if (uint32_t(row["Code"])==1003)
794	format(row["Message"].c_str());
795	else
796	cout << row["Message"] << '\n' << endl;
797
798	}
799
800	}
801	catch (const exception &e)
802	{
803	cerr << '\n' << query << "\n\n";
804	cerr << "SQL query failed:\n" << e.what() << endl;
805	return 12;
806	}
807
808	return 0;
809	}
810
811	// -------------------------- Request data from database -------------------
812	if (verbose>0)
813	cout << "Requesting data... please be patient!" << endl;
814
815	if (verbose>2)
816	cout << '\n' << query << endl;
817
818	const mysqlpp::UseQueryResult res =
819	connection.query(query).use();
820
821	// -------------------------------------------------------------------------
822
823	if (verbose>0)
824	{
825	cout << "Opening file '" << path << "' [compression=" << compression << "]...\n";
826	cout << "Writing data to tree '" << tree << "'" << (nofill?" (--skipped--)":"") << endl;
827	if (num_split)
828	{
829	cout << "Splitting configured " << (split_seq.empty()?"randomly":"in sequence") << " into " << num_split << " branches." << endl;
830	if (!split_quant.empty())
831	cout << "Seed value configured as " << conf.Get<uint64_t>("seed") << "." << endl;
832	}
833	}
834
835	// ----------------------------- Open output file --------------------------
836	TFile tfile(path, update?"UPDATE":(force?"RECREATE":"CREATE"), "Rootify SQL", compression);
837	if (tfile.IsZombie())
838	return 13;
839
840	// -------------------------------------------------------------------------
841
842	// get the first row to get the field description
843	mysqlpp::Row row = res.fetch_row();
844	if (!row)
845	{
846	cerr << "Empty set returned... nothing to write." << endl;
847	return finish(connection, verbose, profiling, print_connection)+20;
848	}
849
850	if (verbose>0)
851	cout << "Trying to setup " << row.size() << " branches..." << endl;
852
853	if (verbose>1)
854	cout << endl;
855
856	const mysqlpp::FieldNames &l = *row.field_list().list;
857
858	vector<double> buf(l.size());
859	vector<uint8_t> typ(l.size(),'n'); // n=number [double], d is used for DateTime
860
861	UInt_t cols = 0;
862
863	// IMPLEMENT FILE SPLITTING!
864	// OpenFile(tree, query)
865	// SetupColumns
866	// WriteRow
867	// CloseFile
868
869	// Ratio[3]: 50%, 20%, 30%
870	// File[x3]: root, cout, fout
871
872
873	// -------------------- Configure branches of TTree ------------------------
874	vector<TTree*> ttree;
875
876	if (num_split==0)
877	ttree.emplace_back(new TTree(tree.c_str(), query.c_str()));
878	else
879	for (size_t i=0; i<num_split; i++)
880	ttree.emplace_back(new TTree((tree+"["+to_string(i)+"]").c_str(), query.c_str()));
881
882	size_t skipat = 0;
883	size_t skipreg = 0;
884	for (size_t i=0; i<l.size(); i++)
885	{
886	const string t = row[i].type().sql_name();
887
888	if (t.find("DATETIME")!=string::npos)
889	typ[i] = 'd';
890	else
891	if (t.find("DATE")!=string::npos)
892	typ[i] = 'D';
893	else
894	if (t.find("TIME")!=string::npos)
895	typ[i] = 'T';
896	else
897	if (t.find("VARCHAR")!=string::npos)
898	typ[i] = 'V';
899	else
900	if (t.find("CHAR")!=string::npos)
901	typ[i] = 'C';
902
903	bool found = false;
904	for (auto pattern=_ignore.cbegin(); pattern!=_ignore.cend(); pattern++)
905	{
906	if (boost::regex_match(l[i], boost::regex(*pattern)))
907	{
908	found = true;
909	typ[i] = '-';
910	skipreg++;
911	break;
912	}
913	}
914
915	if (l[i][0]=='@')
916	{
917	typ[i] = '@';
918	skipat++;
919	}
920
921	const bool use = l[i][0]!='@' && typ[i]!='V' && typ[i]!='C' && !found;
922
923	if (verbose>1)
924	cout << (use?" + ":" - ") << l[i].c_str() << " [" << t << "] {" << typ[i] << "}\n";
925
926	if (use)
927	{
928	// string name = l[i];
929	// for (const auto &m: mymap)
930	// name = boost::regex_replace(l[i], boost::regex(m.first), m.second);
931
932	for (auto it=ttree.begin(); it!=ttree.end(); it++)
933	it[0]->Branch(l[i].c_str(), buf.data()+i);
934	cols++;
935	}
936	}
937	// -------------------------------------------------------------------------
938
939	if (verbose>1)
940	cout << endl;
941	if (verbose>0)
942	{
943	if (skipreg)
944	cout << skipreg << " branches skipped due to ignore list." << endl;
945	if (skipat)
946	cout << skipat << " branches skipped due to name starting with @." << endl;
947	cout << "Configured " << cols << " branches.\nFilling branches..." << endl;
948	}
949
950	// ------------------------- Open the ascii files --------------------------
951
952	vector<ofstream> fout;
953	if (!write.empty())
954	{
955	vector<string> names;
956	if (num_split==0)
957	names.emplace_back(write);
958	else
959	for (size_t i=0; i<num_split; i++)
960	names.emplace_back(write+"-"+to_string(i));
961
962	for (auto it=names.cbegin(); it!=names.cend(); it++)
963	{
964	fout.emplace_back(*it);
965	if (!*fout.rbegin())
966	cout << "WARNING: Writing to '" << write << "' failed: " << strerror(errno) << endl;
967	}
968	}
969
970	// ----------------------- Prepare the ascii comment -----------------------
971
972	string contents;
973
974	istringstream istr(query);
975	size_t line = 0;
976	bool header = true;
977	while (istr)
978	{
979	string ibuf;
980	getline(istr, ibuf);
981	const string sbuf = Tools::Trim(ibuf);
982
983	const bool shabang = line==0 && ibuf[0]=='#' && ibuf[1]=='!';
984	const bool comment = sbuf[0]=='#' && !shabang;
985	const bool isquery = !shabang && !comment;
986	if (isquery)
987	header = false;
988
989	line++;
990
991	if ((copy_shabang && shabang) \|\|
992	(copy_header && comment && header) \|\|
993	(copy_query && isquery) \|\|
994	(copy_comments && comment && !header))
995	contents += '#' + ibuf + '\n';
996	}
997
998	// ----------------------- Write the ascii headers -------------------------
999
1000	ostringstream htxt;
1001	if (display \|\| !fout.empty())
1002	htxt << row.field_list(delimiter.c_str());
1003
1004	if (display)
1005	{
1006	cout << endl;
1007	cout << contents << endl;
1008	cout << "# " << htxt.str() << endl;
1009	}
1010	for (auto ff=fout.begin(); ff!=fout.end(); ff++)
1011	{
1012	*ff << contents;
1013	*ff << "# " << htxt.str() << endl;
1014	}
1015
1016	// ---------------------- Fill TTree with DB data --------------------------
1017
1018	const uniform_real_distribution<double> distribution(0,1);
1019	mt19937_64 generator;
1020	generator.seed(conf.Get<uint64_t>("seed"));
1021	auto rndm = bind(distribution, generator);
1022
1023	size_t count = 0;
1024	size_t skip = 0;
1025	do
1026	{
1027	size_t index = 0;
1028	if (!split_lut.empty())
1029	index = split_lut[count % split_lut.size()];
1030	if (!split_quant.empty())
1031	{
1032	const float r = rndm();
1033	for (; r>=split_quant[index]; index++)
1034	if (index==split_quant.size())
1035	break;
1036	}
1037
1038	count++;
1039
1040	ostringstream rtxt;
1041	if (display \|\| !fout.empty())
1042	rtxt << row.value_list(delimiter.c_str(), mysqlpp::do_nothing);
1043
1044	if (display)
1045	cout << rtxt.str() << '\n';
1046	if (!fout.empty())
1047	fout[index] << rtxt.str() << '\n';
1048
1049	size_t idx=0;
1050	for (auto col=row.begin(); col!=row.end(); col++, idx++)
1051	{
1052	if (!ignorenull && col->is_null())
1053	{
1054	skip++;
1055	break;
1056	}
1057
1058	switch (typ[idx])
1059	{
1060	case 'd':
1061	buf[idx] = time_t((mysqlpp::DateTime)(*col));
1062	break;
1063
1064	case 'D':
1065	buf[idx] = time_t((mysqlpp::Date)(*col));
1066	break;
1067
1068	case 'T':
1069	buf[idx] = time_t((mysqlpp::Time)(*col));
1070	break;
1071
1072	case 'V':
1073	case 'C':
1074	case '-':
1075	case '@':
1076	break;
1077
1078	default:
1079	buf[idx] = atof(col->c_str());
1080	}
1081	}
1082
1083	if (idx==row.size() && !nofill)
1084	ttree[index]->Fill();
1085
1086	row = res.fetch_row();
1087
1088
1089	} while (row);
1090
1091	// -------------------------------------------------------------------------
1092
1093	if (display)
1094	cout << '\n' << endl;
1095
1096	if (verbose>0)
1097	{
1098	cout << count << " rows fetched." << endl;
1099	if (skip>0)
1100	cout << skip << " rows skipped due to NULL field." << endl;
1101
1102	for (size_t i=0; i<ttree.size(); i++)
1103	cout << ttree[i]->GetEntries() << " rows filled into tree #" << i << "." << endl;
1104	}
1105
1106	for (auto it=ttree.begin(); it!=ttree.end(); it++)
1107	(*it)->Write();
1108	tfile.Close();
1109
1110	if (verbose>0)
1111	{
1112	const auto sec = Time().UnixTime()-start.UnixTime();
1113
1114	cout << Tools::Scientific(tfile.GetSize()) << "B written to disk.\n";
1115	cout << "File closed.\n";
1116	cout << "Execution time: " << sec << "s ";
1117	cout << "(" << Tools::Fractional(sec/count) << "s/row)\n";
1118	cout << "--------------------------------------------------------------" << endl;
1119	}
1120
1121	return finish(connection, verbose, profiling, print_connection);
1122	}

Note: See TracBrowser for help on using the repository browser.

Download in other formats: